You’ve read one of your four complimentary articles for this month.
You can read four articles free per month. To have complete access to the thousands of philosophy articles on this site, please SUBSCRIBE!
How To Get Off Our Trolleys
Phil Badger tackles the famous ‘Trolley Problem’ of ethics.
So-called ‘Trolley Problems’, in which we are confronted by tortuously difficult ethical dilemmas, have become part of the stock-in-trade of moral philosophy over the last few years. The basic scenario, which can be endlessly tweaked to provoke ever more rigorous examination of our moral reasoning, concerns a run-away trolley on an imaginary railway line. The trolley is on a lethal collision course with an oblivious group of (say, five) workmen, but can potentially be diverted onto another track by your pulling on a conveniently-placed lever. The problem is that this will place the trolley on a different, equally lethal collision course with an innocent individual.
Readers who haven’t come across this thought-experiment before might pause for a moment to consider what they might do. Some of you will undoubtedly be in favour of hauling on the lever and consigning the poor innocent singleton to oblivion. We can certainly understand why some people might be minded to do so. In the situation presented there seems a clear choice between allowing the deaths of a group of people, and acting to end the life of one. For them the ‘best’ option – meaning the one that will cause the least suffering – is to favour the many. For others the story will not be so simple, and they will draw a distinction between ‘killing’ one individual by active intervention and ‘letting several die’ by remaining passive. This is a distinction we’ll return to later.
Problems With Morality
What we have in this thought-experiment is a place to start thinking about our instinctive moral responses (what philosophers call our ‘moral intuitions’) and the way that these intuitions may or may not cohere together or be capable of any kind of rational justification.
For a so-called ‘act utilitarian’ like Jeremy Bentham (1784-1832), the measure of the moral validity of any particular act is the extent to which it promotes pleasure/benefit and minimises pain/suffering. For Bentham our trolley problem would be a ‘no brainer’: levers would be pulled and our lone protagonist squashed without regret. But an unequivocal utilitarian would be a colder fish than most of us might want to spend any time with. Perhaps one of Mr Spock’s fully Vulcan friends might be an unperturbed ‘trolley diverter’; but the rest of us would be in for a lot of bad dreams. The ‘lever puller’ would deserve our utmost sympathy and support (even if we disagreed with her decision) precisely because we would empathise both with her momentary dilemma and her subsequent anguish.
What this tells us depends on our philosophical perspective. Presumably, Bentham would bemoan our lily-livered sentimentality and remind us of the superior rationality of the utilitarian case. Someone had to die, and it is better, in this case, that it should be some one and not a whole group. Of course this highlights the problem of what we might call the essential ‘relativism’ of the utilitarian position: any act might be permitted and a similar act be prohibited depending on the circumstances. Sometimes it will be acceptable to squash the individual, and sometimes the group (imagine that we recognise the lone individual as a world famous surgeon whom we anticipate will go on to save and improve many lives). Of course, making morality relative to the consequences of actions is, on another level, a morally absolute principle (we must always act to maximise ‘utility’). But this is a principle that most of us find deeply disturbing for the following reason.
Imagine applying our trolley logic to the case of the death penalty. Imagine further that a new study showed that, without question, the death penalty really does cut down the number of murders committed in any given year. Surely, under such (admittedly hypothetical) circumstances, the lever diverting the trolley would be rapidly replaced by the lever operating the executioner’s trapdoor. In fact, the replacement is made easier when we consider that the ‘sacrificed’ individual is likely to be a cold-blooded murderer. The wrinkle here is the word ‘likely’, because, from a purely utilitarian perspective, the occasional execution of an innocent makes no difference to the morality of the death penalty – the net benefit justifies the sacrifice.
At this point some readers might be feeling a little uneasy because they can feel that no slope is slipperier than the one we’re now on. So let’s say that it turns out that executing the family of a murderer is even more likely to produce a net reduction in pain and suffering – perhaps the deterrence effect is so strong that we only need to wipe out one family per year to guarantee a violence-free life for the rest of us.
Or, imagine a further series of trolley-like situations. Firstly, a terrorist group hijacks an airliner and demands that David Cameron be handed over to them for ‘revolutionary justice’. If we do not comply with their demands, they will blow up the airliner and all on board. We might construct a utilitarian argument for not giving in to the terrorists on the grounds that doing so will encourage further outrages of the same kind; but is this the only moral reason not to drag Mr Cameron to his death? Most of us would think not.
Take another situation: a variation on the classic ‘ticking bomb’ problem, in which we are invited to consider the morality of using torture to help discover the location of a device primed to explode with devastating effects on the lives of hundreds or even thousands. We might be sympathetic to the desperation of the field officer who reluctantly gets out his pliers; but what if, to extract the necessary information, they end up being employed on the terrorist’s five-year-old child?
We can’t dodge the ethical issue by claiming that torture doesn’t work, any more than we could by arguing that the death penalty is not an effective deterrent. The point of such thought-experiments is to face up to how utilitarian we actually would be prepared to be when the ‘utility’ of an act is not in question. Does there come a point where we must say that some actions, such as the torture of children, are always wrong, no matter what the net benefit to others may be?
However, again, the answer is not obvious. The problem with the ‘morally absolute’ position is two-fold. Firstly, the consistent utilitarian might insist that we be clear what our limits are to how much avoidable suffering are we prepared to allow in order to keep our hands clean: do we waver when the stakes are sufficiently high? Let’s say torturing the child saves 100,000 lives. Does that make a difference? How about a million? Secondly, what kinds of act do we decide to be beyond the pale?
Shall we reply that morality has nothing to do with consequences, or should we try to find a way in which the consequences can be given a weight, but only so far that they don’t lead to ‘unacceptable’ places? The first position is associated with Kant, and leads to some very odd moral conclusions.
Can & Kant
For Kant, morality was a matter of ‘categorical imperatives’, which are best understood as absolute rules which could be generated by the use of reason by any morally-competent individual. Kant gives three formulations of the categorical imperative which are superficially dissimilar but which turn out to be equivalent. There isn’t space here to do anything like justice to any of the formulations; but, to begin with, we’ll consider his idea that we should only act in such a way as we would will the principle (his word is ‘maxim’) of our action to be a universal law. In other words, we should act only in ways that we would want others to act in similar circumstances.
It is notoriously easy to misread Kant as arguing that the consequences of our actions are what matters in judging their moral rectitude. If we say, as Kant says, that we should only act in ways we wouldn’t mind other people acting, it looks pretty much as if we’re asking ‘what would the consequences be if everyone acted that way?’ But as Michael Sandel has pointed out in Justice (2009), this is not what Kant is saying. In fact, Kant thinks that we need to act towards others in a way that respects what it is in them that I demand is respected by them in me – my capacity for autonomous (self-governing) action. What’s wrong with stealing is not that the world would descend into chaos if everyone did it (this would be a ‘rule utilitarian’ argument), but that in stealing I act without respect for the autonomy of the person I steal from, and so imply that no one’s autonomy matters.
Kant’s view of morality is hugely demanding, and in the words of another of his formulations, asks us to ‘treat others as not a means to an end but as an end in themselves’. Thus for Kantians, utilitarian calculations about the benefits of sacrificing David Cameron, executing the innocent or pulling ‘trolley diverting’ levers are not to be entertained. On this deontological (duty-based) perspective, some things are absolute wrongs.
This position might make us all feel a lot more comfortable than the utilitarian nightmare described earlier; but once again, things are not so easy, because in severing the connection between the morality of an act and its consequences, Kant commits himself to a position many of us find just as difficult.
Famously, Kant argues that to lie is an absolute wrong – how can we respect the autonomy of others if we intentionally deprive them of the truth? This sounds fine until we run into what I’ll call ‘the case of the hidden Jews’. Imagine that you’re hiding a family of Jewish people from the Nazis, and that you have an SS officer at the door asking you if you’ve seen them. Kant does some famous wriggling about such issues, in that he suggests that we’re allowed to fail to give the information (we might tell the SS officer that “there were Jews about at some point in the past”) but not to lie. But this is not very convincing: we all want to say that the moral thing is to lie in the most convincing way possible to save them. In other words, in this and equivalent cases, the terrible consequences of not lying are more important than any abstract principle [see also here].
Attempts at Reconciliation
To say I’ve oversimplified things here might be a bit of an understatement, but the point remains that our ‘moral intuitions’ – our gut instincts – evidently pull us in two directions. At times it seems right that we aim to minimise suffering in a utilitarian style, while, at others, we want to be deontologists (absolutely principled) like Kant. The big question is, can the two positions be coherently reconciled?
If we think a bit more about our David Cameron example, we might get a clue about how to make a start in this direction. One response would be to argue that the terrorists who hijack the plane and threaten to blow it up are responsible, so the terrible consequences of not complying with their demands should not weigh on our consciences. This position, proposed by the neo-Kantian Alan Gewirth in Reason and Morality, (1980), is plausible, but not altogether helpful. One problem with it is that it does not apply to situations in which we cannot blame someone else for our actions. We might suppose that we could try to blame the terrorist for the agonies of his tortured child in the ‘ticking bomb’ case; but very few will be much convinced by this move. Similarly, if we pull the lever in our initial trolley scenario, there can be no doubt that we have chosen to do something and that responsibility rests with us. The same logic applies, for example, to the use of deadly force against civilian populations in war time. Dropping atom bombs might seem justified on grounds of utility – to curtail even more extensive suffering by ending the war – but not on the grounds that the leaders of the other side made us do it.
Another try would be to distinguish between ‘acts’ and ‘omissions’, such that we see ourselves as responsible for the former but not the latter. The most usual employment of this distinction is in distinguishing between ‘killing’ and ‘letting die’ in the case of gravely ill patients nearing the end of life. ‘Killing’ would involve administering drugs to terminate the patient’s life in order to end their suffering, while ‘letting die’ might involve the withholding of treatment in order to hasten death. The problem here is that doing nothing might cause significant amounts of suffering in its own right. Not hydrating or feeding a patient (or not pulling the lever to divert the trolley) might cause more suffering than active intervention. (This is one reason why many philosophers share my sense that the distinction between acts and omissions is inherently dubious.)
Thirdly, we can invoke the so called ‘principle of double effect’, which suggests that we can’t be held responsible for the unintended or secondary effects of particular actions. A terminally ill patient may be given large doses of morphine to alleviate pain, and this might also, incidentally, hasten his death. Or perhaps a leader might believe he can bring an end to some period of terrible human suffering by, for instance, dropping an atomic bomb on a Japanese city. In this case his principle goal is not to kill the maximum number of non-combatants, but rather to save lives in the long run. Similarly, when we pull the lever to divert the trolley, we do not intend the death of the hapless individual who is now in its path, but rather to save the lives of the group who now aren’t. Yet the trolley example is not quite analogous to the atomic bomb scenario, and we have to invoke a famous refinement to make it so. Imagine that we can only save the group by pushing an fat individual into the trolley’s path. Now we have to will the death of the fat man to bring about our goal. This surely feels closer to the situation of nuclear attack that President Truman must have agonised over (ironically, one of the bombs dropped on Japan was codenamed ‘fat man’). Yet by the same logic, we have to acknowledge that killing the innocents is not just a side effect of dropping the bomb, so the principle of double effect does not apply.
Freedom and Sacrifice
Again, we have to face the competing demands of powerful and intuitively plausible understandings of morality. Some things seem wrong no matter what the supposed benefits might be. Torturing children to extract information from the enemy seems to fall under this heading: the ends don’t justify the means. However, the sacrifice of the few for the many is difficult to resist. So if the trolley or the bomb or the hangman’s rope kills a few innocents and saves many, what are we to do? It is easy to be an absolutist when you don’t have Truman’s job, and a true utilitarian will tell us we are just being self-indulgent when we entertain our moral qualms anyway.
At this point let’s consider the psychology of the situation. Firstly, few of us would object to the notion of individual sacrifice when such sacrifice is voluntary. Imagine the fat man hurling himself in the path of the trolley, or Mr Cameron stoically arriving at the airport and, unimpressed by our pleas, handing himself over to the terrorists to face his fate. Under these circumstances, the horror of using a person as a means to an end is transformed to awe at an act of human nobility. Being the beneficiary of such an act would be a huge burden, but less so t han being the beneficiary of the death of an unwilling and terrified individual. This acknowledges our natural revulsion to the horrific injustice perpetrated upon the involuntary sacrifices.
So, where does all this leave us? Can we give a plausible account of morality which takes into consideration both the sense that we have to weigh the consequences of our actions and also the sense that, nonetheless, there are moments when consequences are secondary to higher principles? Perhaps we can, but only at a great cost, which includes substituting Kant’s notion of ‘autonomy’ for a sense of autonomy we’ll find more familiar.
The first point is that autonomy, in the sense of my ability to choose my actions uncoerced by others, doesn’t always matter to us. My freedom is limited by many laws, and yet I don’t feel conspicuously ‘un-free’. In some countries I’ll be fined for crossing a road on a red light, but that doesn’t make me ‘un-free’. What matters supremely is that those things about which I get to make choices are the ones which make life meaningful to me (this is the more familiar, non-Kantian sense of ‘autonomy’). The law doesn’t prescribe my values or my deepest commitments, nor should it. My big life choices must remain matters of non-interference, because robbing me of these things would render my life meaningless to me. Thus, on neo-Kantian grounds, I have to will the same respect for what I call ‘the large-scale concepts of the good’ of others. On this basis, nobody can be handed over to terrorists against their will, unjustly hung, or thrown in front of a runaway trolley. (Although, i n the case of our initial trolley scenario, since there is no possibility of consulting any of the potential victims about what they would choose to do, we might argue that their autonomy becomes irrelevant to the situation and utilitarian considerations become primary. Thus it is right to pull the lever and save the many. We might even decide to ascribe heroic status to the lone victim – after all, she might have decided to sacrifice herself. This is certainly more likely than the whole group of people on the other track doing so for her.)
At this point we might be feeling a little smug because we have, apparently, rendered two plausible but apparently conflicting moral principles compatible. However, I mentioned earlier that there was a cost to my strategy. In making the holding of personal large-scale concepts of the good a criteria for moral status, I’ve opened up two unpalatable possibilities. Firstly, that beings who have no capacity for holding such concepts must be accounted to be morally inferior – which might sound okay until you realise that small children, the mentally ill, the intellectually impaired and animals all come into this category. Secondly, if, for ‘morally immature’ beings, we then make the capacity to suffer the only applicable criteria for their moral consideration, there are no ethical grounds for favouring infant or mentally challenged humans over many other creatures. For some philosophers, such as Peter Singer (Practical Ethics, 1979), this moral equality with animals is not much of a problem – but for others it certainly will be. You might also point out that I seem to have committed myself to the position that a comatose fat man can be dropped onto the tracks without a quibble (we can’t consult him on his large-scale concept of the good, and he’s likely to suffer far less than the fully conscious guy who’ll get quashed if we pull our original lever).
Perhaps, in the end, we have to invoke the collective ‘large-scale concept of the good of society’ (or even humanity) to resolve these remaining issues. Perhaps, to use Rawls’ phrase, we can rely on an ‘overlapping consensus’ about values in society’s concept of large-scale good: no decent human being would want to be saved by information acquired by the torture of a child. Perhaps; but there seems some serious question-begging going on at this point about what’s decent. Yet my hope is that a social consensus on non-negotiable goods is the result of the kind of discussion we’ve been engaged in, rather than being a fairly implausible foundation for it. This is a place to start a conversation and not to end it.
© Phil Badger 2011
Phil Badger teaches philosophy and psychology in Sheffield. He would like to thank Tony Cole for provoking him into writing this article.
The Trolley Problem
Philippa Foot first introduced the Trolley Problem in her 1967 paper ‘Abortion and the Doctrine of Double Effect’. It continues to provoke debate amongst moral philosophers.
Judith Jarvis Thompson first proposed the ‘fat man’ version of the famous dilemma, in which someone can prevent five deaths by hurling a fat man in front of the trolley. Thompson draws a distinction between this scenario and the original one, in which harm is diverted from the group to the individual. The latter case requires deliberately involving the ‘fat man’ in the incident.
On the other hand, are the two scenarios substantively different other, given that one involves a shove and the other a pull on a lever? American philosopher Peter Unger would argue that a fundamental difference is hard to pin down, even if throwing the fat man in front of the trolley appears more cruel and involved. If the fat man is standing next to us, are we not in the same way merely ‘diverting damage’ by hurling him to his death (albeit in a more ugly way)?
An act utilitarian – someone who judges the value of each action according to its net yield of happiness – would sympathise with this view. However, a rule utilitarian, who thinks that laws should be set up in order to increase the net happiness in the long run, may disagree on the basis that if a ‘moral rule’ were created which permitted the hurling of fat people in front of trolleys to save lives, adherence to this rather unjust rule would not maximise net happiness, so the rule should therefore never be sanctioned.